Clang only does lock annotation checks based on annotations that are
already visible, so for best results they need to be on prototypes as
well as definitions.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Each of these functions had only a single caller, and breaking them out as
separate functions didn't seem to many the code clearer.
I added a new function meter_insert_rule(): oftable_insert_rule() had used
members of struct meter directly, which worked because it was after the
definition of struct meter. So there was a choice to either move the
definition of struct meter earlier or add a helper function; I chose the
latter.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Thomas Graf <traf@suug.ch>
I am unsure if it is appropriate to annotate OPENFLOW-1.1+ in this way
but I wish to throw my hat into the ring for work on EXT-187/OF1.4
flow monitors.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
OFPRR_METER_DELETE was introduced in OF1.4 however meters were introduced
in OF1.3.
Regardless of the OF version when flows are deleted cause flows to be
deleted handle_delete_meter() calls delete_flows__() with
OFPRR_METER_DELETE as the reason.
In order to avoid sending OFPRR_METER_DELETE to controllers connected
using OF1.3 map OFPRR_METER_DELETE to OFPRR_DELETE which exists in that
version.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
This is in preparation for using the same helper as part of support
for using recirculation in conjunction series of actions including
with MPLS actions that are currently not able to be translated.
In that scenario the idle timeout will be used to expire internal
rules that are added to handle recirculation.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
This is to allow a recirculation id to be associated with a rule
in the case that its actions cause recirculation.
In such a case if the recirc_id field is non-zero then that value should be
used, otherwise a value should be obtained using
ofproto_dpif_alloc_recirc_id and saved in recirc_id field.
When destructing the rule if the recirc_id field is non-zero then
the associated internal flow should be deleted.
This is in preparation for using the same helper as part of support
for using recirculation in conjunction series of actions including
with MPLS actions that are currently not able to be translated.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
This is in preparation for using this value
in ofproto-dpif-xlate.c when composing recirculation
actions added as a result of processing (MPLS) actions.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
According to msdn documentation, one is discouraged from using
IP_TOS for ipv4 sockets (it apparently does not actually set
anything). Also, IPV6_TCLASS does not work in
Windows (it always returns an error and also is undocumented).
Looks like Microsoft recommends QoS2 APIs to achieve
the same. Till we add those API calls, simply return on Windows.
(Noticed while running unit tests for ipv6. set_dscp would fail.)
Signed-off-by: Gurucharan Shetty <gshetty@nicira.com>
Acked-by: Ben Pfaff <blp@nicira.com>
This further eases porting existing hmap code to use cmap instead.
The iterator variants taking an explicit cursor are retained (renamed)
as they are needed when iteration is to be continued from the last
iterated node.
Signed-off-by: Jarno Rajahalme <jrajahalme@nicira.com>
This bug did not manifest due to 'hmap_node' being in the same offset
in both struct cls_partition and struct cls_subtable.
Signed-off-by: Jarno Rajahalme <jrajahalme@nicira.com>
My reading of handle_delete_meter() is that OFPRR_METER_DELETE is used.
Accordingly delete the entry relating to it from OPENFLOW-1.1+.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
Use OFPRR_GROUP_DELETE as the reason for deleting flows due
to the removal of a group that they use.
This implementation adds an delete_reason member to struct ofputil_flow_mod
as a convenient way to set the reason used by delete_flows__() when it is
called indirectly from delete_group__().
Signed-off-by: Simon Horman <horms@verge.net.au>
[blp@nicira.com initialized the new member in a few more places]
Signed-off-by: Ben Pfaff <blp@nicira.com>
I think these were leftovers from the removal of %z for MSVC that happened
some time ago.
VMware-BZ: 1265762
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Pritesh Kothari <pritesh.kothari@cisco.com>
Ofproto-dpif bond tests relies on netdev-dummy ports to set up
socket connection before actual tests. The time required
for socket connection varies depends on system load, making the test
prone to failure with simple sleep calls. On the other hand,
conservative sleep value for the slowest machine may be excessive on
a faster machine.
This patch removes the sleep calls. Replace them with
WAIT_FOR_DUMMY_PORTS() introduced in the last patch, thus removing
the timing dependency.
CC: Jarno Rajahalme <jrajahalme@nicira.com>
Signed-off-by: Andy Zhou <azhou@nicira.com>
Acked-by: Jarno Rajahalme <jrajahalme@nicira.com>
Using without any parameter, this command list the connection
state of all netdev-dummy devices that are configured to make
active connections.
Optionally, the name of the netdev-dummy device can be supplied
as a parameter.
The states will be displayed as:
"connected": The socket has been connected to the listener.
"disconnected": The socket is not connected.
"unknown": It is not an active dummy device.
CC: Jarno Rajahalme <jrajahalme@nicira.com>
Signed-off-by: Andy Zhou <azhou@nicira.com>
Acked-by: Jarno Rajahalme <jrajahalme@nicira.com>
Added details of dpdk poll mode setup to make it easier
for someone, not familiar, to get it operating.
Signed-off-by: Mike A. Polehn <mike.a.polehn@intel.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Skip some of the tests that uses '--monitor' as an option as
it is not implemented on Windows.
When a 'kill pid' is done on windows (through 'taskkill //F'),
pidfiles are not deleted (because it is force kill), so use
'ovs-appctl exit' instead.
Signed-off-by: Gurucharan Shetty <gshetty@nicira.com>
Acked-by: Ben Pfaff <blp@nicira.com>
Fix the bug which did not kill the processes with pids that had
a zero in them. Also, some tests do a AT_CHECK([kill `cat pid`]) which
on windows prints something on the stdout causing the tests to fail.
So supress it.
Signed-off-by: Gurucharan Shetty <gshetty@nicira.com>
Acked-by: Ben Pfaff <blp@nicira.com>
With Visual Studio and Msys combination, something is going
wrong when we do a '3<pcap' followed by fdopen(3, 'rb'). fdopen
fails. I do not know the exact reason for the failure. But the
workaround is straightforward.
Signed-off-by: Gurucharan Shetty <gshetty@nicira.com>
Acked-by: Ben Pfaff <blp@nicira.com>
On some systems libintel_dpdk.a fails to link with libopenvswitch
unless -ldl is used. This should address the issue
Signed-off-by: Daniele Di Proietto <ddiproietto@vmware.com>
Signed-off-by: Ben Pfaff <blp@nicira.com>
netflow_flow_clear() asserted that no packets or bytes were included
in the statistics for the flow being cleared. Before threading Open
vSwitch, this assertion was always true because netflow_expire() was
always called before calling netflow_flow_clear(). Since Open
vSwitch was threaded, however, it was possible that a packet arrived
after netflow_expire() but before netflow_flow_clear(), since each of
these function separately took the netflow mutex.
This commit fixes the problem by merging netflow_expire() into
netflow_flow_clear(), under a single acquisition of the netflow
mutex.
Signed-off-by: Anoob Soman <anoob.soman@citrix.com>
[blp@nicira.com modified the patch to remove netflow_expire() and
rewrote the commit message]
Signed-off-by: Ben Pfaff <blp@nicira.com>
Most other code in Open vSwitch that works with flow cookies keeps them
in network byte order. Using network byte order in struct ofpact_learn,
also, reduces the number of byte order conversions needed across the
source tree.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
This flag was overlooked when support for the "learn" action was added.
(It was always supported in the OpenFlow code, just not in the ovs-ofctl
interface.)
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
recirc_id_pool_create() allocates memory but recirc_id_pool_destroy() did
not destroy it.
Found by valgrind.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Rather than setting and resetting the 'mark' field in the ukey, this
patch introduces a seq to track whether a flow has been seen during the
most recent dump. This tidies the code and simplifies the logic for
detecting when flows are duplicated from the datapath.
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Acked-by: Ethan Jackson <ethan@nicira.com>
Without this change a monitor update will be sent when a flow mod changes
a rules cookie but not if only the actions are updated. This appears
to be a logic error.
I noticed this while working on implementing OpenFlow1.4 flow monitor
as an OpenFlow1.4 flow mod does not update a rules cookie.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Ben Pfaff <blp@nicira.com>
This is partly documentation of how patches are tagged in practice in Open
vSwitch. The bits at the end about "Reported-at:" and "VMware-BZ:" are
an attempt to influence future practices; I cannot say how successful they
will be.
I am not sure whether these key-value pairs at the end of commit messages
are actually commonly called "tags". I'm happy to use a different term if
that one seems wrong.
Reported-by: Flavio Leitner <fbl@redhat.com>
Reported-at: http://openvswitch.org/pipermail/dev/2014-June/040952.html
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Flavio Leitner <fbl@redhat.com>
Acked-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
'wevent' isn't actually used on non-Windows systems, but poll_fd_wait_at()
and find_poll_node() treat events with different 'wevent' as different, so
it seems better to make sure that 'wevent' doesn't matter.
Alternatively, one could ovs_assert(!wevent). I guess that would catch
a caller accidentally swapping the 'fd' and 'wevent' arguments.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Gurucharan Shetty <gshetty@nicira.com>
Otherwise the indeterminate 'wevent' could frustrate poll_fd_wait_at()'s
attempt to merge "poll_node"s for the same fd.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Gurucharan Shetty <gshetty@nicira.com>
flow_compose_l4() can cause 'b' to be reallocated, thus the network header
pointer needs to be refreshed afterward.
Found by valgrind in the IPv6 case. I updated the IPv4 case too just in
case, and for consistency.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Andy Zhou <azhou@nicira.com>
Commit 87400a3d4c (dpif-netdev: Fix use-after-free in port_unref().)
fixed one use-after-free in the common case of port_unref(). However,
there was another, similar case: if port->netdev has no rxqs, then
the netdev_close() causes port->netdev to be destroyed and thus the
following call to netdev_n_rxq() accesses freed memory. This commit fixes
the problem.
Found by valgrind.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Andy Zhou <azhou@nicira.com>
The dp_hash and recirc_id fields weren't being initialized.
Found by valgrind.
Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Andy Zhou <azhou@nicira.com>
Add 2 second sleep before time stop to allow internal ports to
reconnect in case the initial connection attempt failed.
Bond tests sometimes persistently fail without these delays.
Signed-off-by: Jarno Rajahalme <jrajahalme@nicira.com>
Acked-by: Andy Zhou <azhou@nicira.com>
One of the reasons that xlate_cache was introduced was to ensure that
statistics were attributed to the correct rules and interfaces according
to the flow that was installed into the datapath, rather than according
to the current state of the flow table.
This patch makes the revalidators use the xlate_cache to attribute stats
when full revalidation is required, as the statistics belong to the old
ruleset. However, when revalidating, the rules may have changed while
leaving the datapath flows intact, so we still re-create the xcache at
that point.
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Acked-by: Alex Wang <alexw@nicira.com>
This patch refactors the code around ukey creation and lookup to
simplify the code for callers. A new function ukey_acquire() combines
these functions and attempts to acquire a lock on the ukey. Failure to
acquire a lock on the ukey is usually a sign that another thread is
handling the same flow concurrently, which means the flow does not need
to be handled anyway.
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Acked-by: Ethan Jackson <ethan@nicira.com>
It's unclear why these tests are named "ofproto-if..." unlike all of the
other ofproto-dpif tests. Rename them to be more consistent.
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Acked-by: Simon Horman <horms@verge.net.au>
When a bridge of datatype type netdev receives a packet, it
copies the packet from the NIC to a buffer in userspace.
Currently, when making an upcall, the packet is again copied
to the upcall's buffer. However, this extra copy is not
necessary when the datapath exists in userspace as the upcall
can directly access the packet data.
This patch eliminates this extra copy of the packet data in
most cases. In cases where the packet may still be used later
by callers of dp_netdev_execute_actions, making a copy of the
packet data is still necessary.
This patch also adds a dpdk_buf field to 'struct ofpbuf' when
using DPDK. This field holds a pointer to the allocated DPDK
buffer in the rte_mempool. Thus, an upcall packet ofpbuf
allocated on the stack can now share data and free memory of
a rte_mempool allocated ofpbuf.
Signed-off-by: Ryan Wilson <wryan@nicira.com>
Acked-by: Jarno Rajahalme <jrajahalme@nicira.com>
Acked-by: Pravin B Shelar <pshelar@nicira.com>
This patch makes sure that the tx and rx queues are allocated on the NUMA socket
chosen at device initalization time, instead of the NUMA socket 0.
Signed-off-by: Daniele Di Proietto <ddiproietto@vmware.com>
Acked-by: Pravin B Shelar <pshelar@nicira.com>
As per netdev-provider interface, netdev_dpdk_rxq_recv should receive at most
NETDEV_MAX_RX_BATCH.
Signed-off-by: Daniele Di Proietto <ddiproietto@vmware.com>
Acked-by: Pravin B Shelar <pshelar@nicira.com>
In case DP_HASH and RECIRC actions need to be executed in slow path,
current implementation simply don't handle them -- vswitchd simply
crashes. This patch fixes them by supply an implementation for them.
RECIRC will be handled by the datapath, same as the output action.
DP_HASH, on the other hand, is handled in the user space. Although the
resulting hash values may not match those computed by the datapath, it
is less expensive; current use case (bonding) does not require a strict
match to work properly.
Reported-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Andy Zhou <azhou@nicira.com>
Acked-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
This patch also checks the system platform as clock_gettime
could exist on different platforms but with different values of
CLOCK_MONOTONIC and different definitions of 'struct timespec'.
In this case, the system call would be expected to catch the
error, which is dangerous.
This patch ensures Linux, NetBSD and FreeBSD platforms use
clock_gettime with their corresponding correct values and
definitions. All other platforms use time.time().
Signed-off-by: Ryan Wilson <wryan@nicira.com>
Acked-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
